Available online at www.CivileJournal.

org

Civil Engineering Journal

Vol. 2, No. 10, October, 2016

Investigation of Ready Mixed Concrete Transportation Problem

Using Linear Programming and Genetic Algorithm
Gulcag Albayrak a*, Ugur Albayrak a
a
Department of Civil Engineering, Eskisehir Osmangazi University, Eskişehir, Turkey.

Received 26 September 2016; Accepted 28 October 2016

Abstract
Ready-mixed concrete (RMC) is one of the most common building material for construction industry for nearly all
developed and developing countries. Generally, because of the technical requirements, concrete must be mixed in a batch
plant and transported to the construction site. There are two important factors affected the cost of RMC: raw material
cost and transportation cost. Additionally, transportation cost is also included when determining the unit price of RMC.
However, profitability affects adversely in the case of long distance between the plant and construction site. For these
reason, distribution of RMC from supply to the demand points with minimum cost is aimed in this study. This work
contributes to both modelling and dispatching of RMC as an optimization problem by applying linear and heuristic
methods. For this purpose, as an example, an urban area which divided into 7 districts and contained 4 concrete batch
plants is discussed. Linear programming and genetic algorithm were applied to solve this problem and compared each
other under the same conditions. The result shows linear programming is more efficient for this application because of
the limited constraints and variables.
Keywords: Optimization; Genetic Algorithm; Mathematical Modeling; Ready-Mixed Concrete.

1. Introduction
Concrete is one of the most widely used construction material in the world which was first introduced into the
construction industry in the early 20th century. Nowadays, concrete is especially prepared by concrete batch plant
according to varied recipes to meet the requirements of the construction site and then transported to the construction
site. Therefore this product is called Ready Mixed Concrete (RMC). To expand the service of RMC without
establishing high cost batch plant at the construction site, the RMC truck was invented to deliver RMC to the
construction site. From the business point of view, the RMC batch plant manager may want to dispatch RMC trucks to
different construction sites as many as possible to maximize the production and profits of the plant. In addition to, the
necessity of the concrete placement in a limited time supports the importance of distance. RMC usually needs to be
cast within 1.5 hour after being produced by the batch plant, which limits the service areas of the RMC batch plant. In
addition, because of limited service area, the business competition between RMC batch plants is intense [1].
Optimization is the first technique that comes to mind to solve RMC transportation problems. There are many
efficient and flexible models and researches based on optimization techniques in literature that have been proposed to
solve RMC production and delivery problems. Especially, in the last decade a number of studies have been conducted
using heuristic methods to obtain notable solutions for this problem by researcher. The majority of them are related to
scheduling. Lu and Lam [2] focused on how to simultaneously optimize concrete delivery schedules and resource
provision for the plant, based on a simulation modeling platform which incorporated a combined discrete event
simulation and genetic algorithms to model and further optimize one-plant-multisite RMC plant operations.
Feng and Wu [3] analyzed factors that impact on the delivery process, then built a model based on genetic
algorithms and simulation techniques to find the best dispatch schedule that minimized the total waiting time for trucks

* Corresponding author: galbayrak@ogu.edu.tr

491
Civil Engineering Journal Vol. 2, No. 10, October, 2016

at a site, while satisfying the need for deliveries to other sites without interrupting the concrete casting operation.
Graham, Forbes and Smith [4] presented a neural network methodology to model the RMC delivery system and found
that two and three-layer feed-forward networks provide the best estimates of concrete placing productivity. Naso,
Surico, Turchiano and Kaymak [5] developed a detailed model and proposed a metaheuristic approach based on a
hybrid genetic algorithm combined with constructive heuristics. Yan, Lai and Chen [6] developed an integrated model
that combined RMC production schedules and truck dispatch in the same framework. The model was formulated as a
mixed integer network flow problem with side constraints, which was solved by developing a solution algorithm
incorporating a mathematical programming solver. Lin, Wang, Huang and Wang [7] provided a detailed formulation of
a multi-objective model of the minimal cost of dispatching RMC trucks. Maghrebi and Waller [8] formulated the RMC
optimization problem mathematically with two models: Integer programming (without time windows) and mixed
integer programming (with time windows). The results of the integer programming and mixed integer programming
models were compared with experts’ decisions.

2. Method
The objective of this paper is to optimize the delivery of RMC. To achieve this goal, these steps are performed:
 Designing the problem as a transportation problem
 Determination of constraints and decision variables
 Building a mathematical model which is appropriate with transportation problem
 Analyzing the model with linear programming and genetic algorithm respectively using Microsoft Solver
Foundation.
The transportation problem is a classic operations research problem where the objective is to determine the
schedule for transporting goods from source to destination in a way that minimizes the shipping cost while satisfying
supply and demand constraints [9].
An ordinary transportation problem has the following components: sources, destinations and weighted edges.
In this paper the transportation problem that we would like to take as an example to compare the outcome of two
optimization methods have a set of nodes called sources, which have a commodity available for distribution, and
another set of nodes called demand points, which require this commodity. The amount of commodity available at each
source and the amount required at each demand point are specified, as well as the cost (especially distance for this
paper) of transporting the commodity from each source to each demand point. The problem is to determine the
quantity to be transported from each source to each demand point, so as to meet all the requirements at minimum total
transportation cost. The commodity of this problem is RMC, the sources are found at concrete batch plants, where the
RMC is produced, and the demand points are the construction sites.

2.1. Optimization Model Design

The transportation problem can be usually described using following mathematical model. In particular, it is
assumed that the transportation network is given by a set Ā of arcs, where (i, j) Ā means that there exists route
connecting the provider i and the demander j.
We denote by cij the unit shipment cost on the arc (i; j), by si the available supply at the provider i, and by d j the
demand at the demander j. The decision variables are the quantities x ij of the commodity that is shipped over the arc (i;
j) Ā and the problem is to minimize the transportation costs,
Minimize ∑( )

Under the natural constraints:

 The supply si at i should exceed the sum of the demands at all j such that (i; j) Ā, i.e.
∑ ( ) ≤ for all i,

 The demand dj at j must be satisfied in the sense that is less or equal the sum of the supplies at all i such that
(i; j) Ā, i.e.
∑ ( ) ≥ for all j,
and
≥ 0 for all i and j

492
Civil Engineering Journal Vol. 2, No. 10, October, 2016

A transportation problem will have feasible solutions if and only if:

∑ =∑

2.2. Genetic Algorithm

Genetic algorithm is a heuristic evolutionary optimization algorithm based on the Darwin’s principle “Survival of
the fittest” [10]. Genetic Algorithm (GA) is initiated in 1975 by Holland [11], which is based on the mechanics of
natural selection and genetics to search through decision space for optimal solution. There are six components in GA
including chromosome structure, initial population, fitness function, selection, genetic operators and termination
condition.
GA is started with a set of solutions (represented by chromosomes) called population. Solutions from one
population are taken and used to form a new population. The new population will be better than the old one. Solutions
which are selected to form new solutions (offspring) are selected according to their fitness value. If their fitness value
is more suitable, then they have more chances to reproduce. This is repeated until some condition (e.g. number of
populations or improvement of the best solution) is satisfied. This algorithm can be represented as Figure 1.

Start

Initialization

Fitness Evaluation

Selection

Crossover

Mutation

Fitness Evaluation

Replacement

No
Termination of GA?

Yes

Stop

Figure 1. A general flow chart of GA

2.3. Numerical Example

A previous study, Albayrak and Albayrak [12] was used for the data of the example. To make the simulation
realistic, a metropolitan area consisting of 7 districts is selected according to this article. In this area, there are 4 batch
plants that supply concrete for all 7 districts. Concrete demands and production capacity of concrete batch plants are

493
Civil Engineering Journal Vol. 2, No. 10, October, 2016

created on the basis of actual values which can be reached in a city of Turkey. The annual demand d j involves all types
of concrete. In practical, the distance (km) from the batch plant to the center point of regions is described as follows
(see Table 1).
Table 1. Distance between suburbans and plants

Suburbans
1 2 3 4 5 6 7
1 15 11 10 10 7.5 6.5 5
Concrete 2 20 16 13 13.5 11.5 8.5 5
Batch Plants 3 8.5 5 1.5 6.5 3.5 7.5 8
4 12.5 8 12.5 6.5 8 13.5 12

The example has its own systematic approach which breaks down the model to four parts: input parameters,
decision variables, constraints and system out.
Decision Variables
xij refers amount of concrete transported from supply point i to demand point j. Assume that i represents the
concrete batch plant (1, 2, 3, 4) and j represents the suburban (1, 2, 3, 4, 5, 6, 7).
Constraints
The production quantity of the concrete batch plant is indicated by si and dj denotes the demand of district. In real
life, the total production capacity of plant is not usually equal to total demand. Therefore, production quantity can be
considered as equal to demand or a dummy demand point can be added to the transportation problem. In this example,
number 8 is a dummy destination. Also, as expected, decision variables cannot be negative values. Thus, that is a
constraint too. Capacities of the concrete batch plants and demands of districts are shown in Table 2 and Table 3.
Constraints are described as follows:
∑ ≤
∑ ≥
xij ≥ 0

Table 2. Capacities of the concrete batch plants

Concrete batch plants

1 2 3 4
Capacities (si)
1050 550 800 600
(m3 x1000/year)

Table 3. Demands of the suburbans

Suburbans
1 2 3 4 5 6 7

Demands (dj)
550 300 400 350 450 150 300
(m3 x1000/year)

Objective function
The purpose is to build a model that covers all demands with the lowest total cost from the sources under the
current constraints. Optimizing the quantity of dispatched concrete is important in this problem. Therefore, concrete
distribution cost is taken into account as $1 per cubic meter in terms of being a unit value.
Minimize ∑( )

494
Civil Engineering Journal Vol. 2, No. 10, October, 2016

3. Results
Primarily, this problem is solved with classical transportation method based linear programming (LP). The
equations of the problem are transferred to Excel Solver Foundation, and then solution of the problem is performed.
An optimum solution is found which covers all the constraints of demands and sources. The optimum amount of
concrete transported from supply point i to demand point j can be shown from Table 4.
All demands are found in the greatest possible value i.e. all demands are fully met. In the same way, all source
values are calculated at the lowest value. However, plant 1 and 2 are observed that they have excess capacity. The
excess capacity (500.000 m3) is assigned number 8 due to use of dummy demand point. The transportation cost of
dummy demand is considered as zero. According to results, total cost of the delivering of RMC is $16.925.000 for a
year.
Table 4. Results of decision variables (xij ) of LP

x11 x12 x13 x14 x15 x16 x17 x18

150 50 - - 450 150 - 250

x21 x22 x23 x24 x25 x26 x27 x28

- - - - - - 300 250

x31 x32 x33 x34 x35 x36 x37 x38

400 - 400 - - - - -

x41 x42 x43 x44 x45 x46 x47 x48

- 250 - 350 - - - -

In addition to, linear programming, GA is built to balance the production of the construction sites and the batch
plants. The population size and mutation rate are defined respectively 100 and 0.075. Likewise, all demands are fully
met, but the outcomes are different. Total cost of the delivering of RMC is calculated s $20.050.000.The values of the
decision variables are given below (Table 5).

Table 5. Results of decision variables (xij) of GA

x11 x12 x13 x14 x15 x16 x17 x18

- - - - 100 150 300 500

x21 x22 x23 x24 x25 x26 x27 x28

- - - 200 350 - - -

x31 x32 x33 x34 x35 x36 x37 x38

- 250 400 150 - - - -

x41 x42 x43 x44 x45 x46 x47 x48

550 50 - - - - - -

4. Conclusion
The concrete is mixed just-in-time before the loading of the vehicle at the plant, or the raw materials are poured
into the vehicle and are mixed on the way to the construction site. In brief, concrete is a perishable good which cannot
be stored or produced in advance. Moreover after being blended the good cannot be transported for more than 1.5
hours [12]. Therefore, transportation algorithm which emphasis on the point of the lowest delivering cost is used in
this paper.
The described model and solution method are sufficiently flexible to be enhanced for typical real world
requirements. At the same time, practical application of the basic model shows that because of the central management
of RMC is more advantageous than the traditional way.
In this paper, the mathematical model which is built for optimization is used. Both of techniques satisfy the balance
between the supply and demand of RMC. However, LP obtained the exact solution and the minimum value for

495
Civil Engineering Journal Vol. 2, No. 10, October, 2016

delivering RMC. The difference between these two methods is that GA does not assure the optimum solution for all
problems. In contrast to LP, GA is capable of searching the optimal value instead of the optimum. In heuristic
methods, even though the results achieved by these techniques are acceptable, there are still many problems in
mathematical modelling that threaten the reliability of the techniques. Also the heuristic methods are powerful tools,
especially the problems have uncertainties.

5. References
[1] Feng, C. W., Cheng, T. M. and Wu, H. T., Optimizing the schedule of dispatching RMC trucks through genetic algorithms,
Automation in Construction, 13 (3). (2004), 327–340.
[2] Lu, M. and Lam, H. C., Optimized concrete delivery scheduling using combined simulation and genetic algorithms, Proceedings
of the 37th conference on Winter Simulation Conference. Orlando. USA, 2005.
[3] Feng, C. W. and Wu, H. T., Integrating fmGA and CY-CLONE to optimize the schedule of dispatching RMC trucks,
Automation in Construction, 15 (2), (2006), 186–199.
[4] Graham, L. D., Forbes, D. R. and Smith, S. D., Modeling the ready mixed concrete delivery system with neural networks,
Automation in Construction, 15 (5), (2006), 656–663.
[5] Naso, D., Surico, M., Turchiano, B. and Kaymak, U., Genetic algorithms for supply-chain scheduling: a case study in the
distribution of ready-mixed concrete, European Journal of Operational Research, 177 (3), (2007), 2069–2099.
[6] Yan, S., Lai, W. and Chen, M., Production scheduling and truck dispatching of ready mixed concrete, Transportation Research.
44 (1). (2008), 164–179.
[7] Lin, P. C., Wang, J., Huang, S. H. and Wang, Y. T., Dispatching ready mixed concrete trucks under demand postponement and
weight limit regulation, Automation in Construction, 19, (2010), 798–807.
[8] Maghrebi, M. and Waller, S. T., Assessing the accuracy of expert-based decisions in the dispatching of ready mixed concrete,
Journal of Construction Engineering and Management, 140(6), (2014), 06014004-1 - 06014004-7.
[9] Aditya, S., Nideesh, K. T. and Guruprasad, N., Solving the transportation problem for warehouses and dealers in Bangalore
City, World Academy of Science, Engineering and Technology Mathematical and Computational Sciences, 2(1), (2015).
[10] Chandwani, V., Agrawal, V. and Nagar, R., Modeling slump of ready mix concrete using genetic algorithms assisted training
of Artificial Neural Networks, Expert Systems with Applications, 42(2), (2015), 885-893.
[11] Holland, J., Adoption in natural and artificial systems, Ann Arbor, MI: The University of Michigan Press, 1975.
[12] Albayrak, G. and Albayrak, U., Optimization of the ready-mixed concrete delivery system using transportation algorithm,
CRC Press, Engineering Optimization IV, The Book Chapter of Numerical Optimization Techniques, 2015, 159-165.

496